What Makes a Solution ‘Implementable’ in the Real World?
At one of our recent INFORMS student meetings at the University of Florida, we hosted a talk by a professor in the school of Industrial and Systems Engineering. This professor has worked with real clients on real projects, and had a lot to share in terms of his experiences working in the fields of analytics and operations research.
But while I, a member of the audience, was entranced with anecdotes of his magnificently simple solutions to seemingly complex problems, I couldn’t help but feel a little less enthralled when he described all the obstacles he faced in actually getting his hard work implemented.
That’s to say, someone could work for months on end in this field on optimization problems ranging from railroad timetables to radiation therapy, and still have very few solutions actually used and implemented by the people who wanted the results in the first place.
So the question I asked myself: what qualities must an analytics solution have to actually be recognized and accepted?
Surely, the quality of the work is at the forefront of this question. Is the work spectacular? If yes, then that’s the first step in actually having an implementable solution.
But when strictly referring to top professionals and thinkers in the field, it was at first difficult for me as a student to grasp the idea that every single hard-thought solution won’t necessarily be used.
And here’s what I realized:
Maybe some industries and some people are still not ready for the efficiency that analytics provides.
Right? Because although analytics and efficiency are by no means novel concepts, their reach is more wide-spread than at any time in the past. While the general concepts aren’t new, the modern implementations of analytics are ground breaking. We’re seeing analytics being applied where you would never think of it applying.
Therefore, for a solution to be implementable, maybe the industry and people who are using the solution’s implementations must be ready to accept this new-found efficiency, even though it may flip the status quo upside down and possibly even change the essence of many people’s job roles in this post-analytics world.
An example that the aforementioned professor gave was that of a classic scheduling problem. He developed an innovative way to automate the scheduling of the daily workings of a factory, from the machine processes to the worker schedules – a fairly common application of OR. But, as he described it, when he went to present his work to the factory representative/advisor, his findings were shot down under the notion that “things have been done the same way here for years, and some computer solutions just can’t replace the work and results of a human working by hand.” Disregarding any delving into the ‘humans vs. computers’ debate, this situation seems to underline the fact that getting solutions implemented in the ‘real world’ takes possibly just as much effort as the work itself. When I began to understand that efficiency can sometimes equate to automation taking over the work of a living human being, the goals of analytics become murkier and less defined.
It seems truer than ever that people are static beings and change isn’t easy. This seems especially true in the analytics world, where the conclusions being drawn are more drastic.
Does this mean that the computer-optimized solution was more efficient – more right – than the factory worker’s hand-written one? Mathematically speaking, yes, assuming the solution is well proven. But things aren’t always so black and white when it comes to real people being affected by complex solutions and results. Things are more grey, more cloudy.
So the optimized solution in this factory situation was completely thwarted.
But the professor closed his talk with some advice that I am grateful to have heard: the people who are going to be affected by your solution must be on your team – on analytics’ side. The people who don’t think in terms of mathematical optimization and number theory are the ones who will feel the rippling effects of analytics. They must be convinced that a solution is more than just numbers. A solution must fit in that grey area, and most of all I think a solution must be flexible and adaptable to what has been coined as the “Human Element.”
While it’s true that people are resistant to change in their everyday lives, I am optimistic that people are willing to work with analytics more now than ever. And that’s great news for everyone. All in all, I think it takes a mutual understanding and respect from each side of the analytics equations for efficiency and change to take place.